Method and apparatus for decoding a video signal using adaptive transform

Abstract

The present invention relates to a technique for encoding and decoding an image, and a method for encoding the image, according to the present invention, comprises the steps of: performing a prediction with respect to a current block; and transforming a residual which is generated by the prediction, wherein in the step of transforming, a first transform is performed on the residual, and then a second transform is performed on a lower frequency area, and wherein a mode of the second transform can be decided on the basis of a mode of the first transform.

The invention claimed is: 
     
       1. A method of decoding a video signal having a current block to be decoded with a decoding apparatus, comprising:
 obtaining residual coefficients relating to the current block from the video signal; 
 obtaining inverse-quantized residual coefficients by inverse-quantizing the residual coefficients; 
 determining, based on transform skip mode information specifying a transform skip mode of the current block, the transform skip mode of the current block among transform skip mode candidates, 
 wherein the transform skip mode candidates include at least one of a 2-dimensional transform mode, a horizontal transform mode, a vertical transform mode and a non-transform mode, and 
 wherein a number of the transform skip mode candidates available to be selected as the transform skip mode of the current block is different based on a partition size of the current block, the current block being a 2N×2N or a N×N partition of a coding block; 
 obtaining, based on the determined transform skip mode, residual samples by selectively performing an inverse-transform for the inverse-quantized residual coefficients of the current block; 
 obtaining prediction samples of the current block; 
 decoding the current block by using the residual samples and the prediction samples; and 
 applying a deblocking filter to the decoded current block, 
 wherein when the inverse-transform is performed for the current block, the obtaining of the residual samples comprises:
 determining, based on a prediction mode relating to the current block and a size of the current block, whether a transform type of the current block is a discrete cosine transform (DCT) or a discrete sine transform (DST), the prediction mode of the current block being inter prediction or intra prediction; and 
 performing the inverse-transform for the inverse-quantized coefficients of the current block by using the determined transform type, 
 
 wherein when the non-transform mode is determined to be the transform skip mode of the current block, the residual sample is obtained by scaling the inverse-quantized residual coefficient with a pre-determined value, 
 wherein the inverse-transform is performed by using one of the discrete cosine transform (DCT) and the discrete sine transform (DST), and 
 wherein the discrete sine transform (DST) is used when the prediction method is intra prediction and the size of the current block is a 4×4 block. 
 
     
     
       2. The method of  claim 1 , wherein the transform skip mode information is obtained based on a size of a transform unit of the current block. 
     
     
       3. The method of  claim 1 , wherein when the prediction samples are obtained using intra prediction, the coding block is partitioned into plural prediction blocks for the intra prediction only if the coding block is a minimum coding block. 
     
     
       4. The method of  claim 3 , wherein the intra prediction is performed on a transform block basis when a size of a prediction block is different from a size of a transform block. 
     
     
       5. The method of  claim 1 , wherein when the prediction samples are obtained using intra prediction, the method further comprises:
 obtaining reference samples of a current block based on neighboring samples adjacent to the current block; and 
 determining whether to apply a smoothing filter to the reference samples based on an intra prediction mode of the current block. 
 
     
     
       6. The method of  claim 5 , wherein when the smoothing filter is applied to the reference samples, one of a plurality of smoothing filters is selected based on a filter information signaled from the video signal.